Page 1
April 2005
An Active Introduction to Academic Misconduct & the Measured Demographics of Misconduct
Fintan Culwin Faculty of Business, Computing and Information Management
London South Bank University Borough Road
London SE1 0AA
Email: [email protected] Telephone: +44 20 7815 7434
Fintan Culwin is a Professor of Software Engineering Education at London South Bank University, which he joined fifteen years ago after a career as a secondary school computing teacher. He has authored six programming text books, holds his PhD in software engineering education and has current interests in forensic document analysis, academic misconduct, automated assessment and learning objects for initial software development education.
Page 2
Abstract
An activity designed to actively introduce first year students to issues
related to academic misconduct is described. The activity involves
computing students writing a short essay on a topic related to the
history of computing. The essays are subsequently automatically
checked for non-originality and the outcomes made available to the
students. The results of approximately 350 students from two different
sessions are analysed for any demographic influences. The only
significant findings were that non-originality was a predictor of non-
completion of first year studies and of a lower percentage outcome.
Page 3
Introduction
The best practice guidance for managing academic misconduct as
summarised by Carroll (Carroll & Appleton 2001, Carroll 2002) includes
the recommendation that all students are explicitly educated at the
outset of their higher education about what constitutes misconduct,
why it is unacceptable and the institutional processes for managing it.
Much of the general advice on how to accomplish this (Culwin &
Lancaster 2001, Carroll 2002, Harris 2004) seems to confuse the
issues of: educating students per se, detecting non-original content in
student work, institutional management of plagiarism and academic
misconduct, and designing assessment processes that are more
resistant to misconduct.
The section of Carroll’s good practice guide titled ‘Active learning
methods to teach students’ (Carroll 2001 Ch 4.3) is only two
paragraphs in length and offers little in the way of detailed suggestion,
but recommends unspecified active learning techniques as most
effective. The general advice given elsewhere regarding educating
students seems largely to consist of informing students as they are
inducted into higher education as to the issues involved and assuming
that thereafter this knowledge will be understood and available. The
availability of the information is guaranteed by the institution hosting it
somewhere in its web presence, and advertising this location to the
student body.
Page 4
However the processes of acculturalisation into higher education need
to embed values related to the ownership of intellectual property in
order to foster attitudes that will respect the importance of correct
attribution of non-original material. It seems unlikely that passive
instruction and abstract discussion will embed such values into a
‘napster generation’ (Barbrook, 2002) who do not accept the legitimacy
of intellectual property rights relating to popular music. This attitude
can be expected to generalise to textual, and other forms of media,
that are Web hosted. An attitude that is given some academic
credibility by some postmodernist theorists (Ross 2004) who deny the
concept of authorship and hence plagiarism. The resulting behaviour in
students is evidenced by the extent of non-original content in submitted
work (Dick et. al 2003) despite the now ubiquitous required student
signature upon submission confirming that the material is original or
correctly attributed.
More active approaches to educating students involve explicitly
illustrating the acceptable and unacceptable ways in which non-original
material can be included in student writing (Culwin 2005). Providing
case studies of plagiarism in practice, quizzes and interactive on-line
learning objects relating to plagiarism (Szondi 2004, Boling & Theodore
Undated). Having the students obtain an essay from an essay bank
and submitting it to a non-originality detection service (Weller 2003).
However there seem to be very few such examples in the general
literature.
Accordingly the first part of this paper describes an activity in which
first year computing students at London South Bank University (LSBU)
Page 5
participated during the 2002/3 and 2003/4 academic sessions. The
activity was intended to provide a positive learning experience,
allowing those who engaged in misconduct to be identified for
educative rather than punitive attention. A by-product of these activities
was direct measurement of the degree of non-originality in a corpus of
student work. This data was subsequently demographically analysed in
an attempt to obtain empirical evidence to support various conclusions
from existing studies based upon less direct evidence.
The UK study that initiated the current interest in plagiarism and
academic misconduct was conducted in 1995 by Franklyn-Stokes and
Newstead (Franklyn-Stokes & Newstead 1995) and, significantly, was
a questionnaire survey. As such it did not attempt to measure
behaviour or the attributes of an artefact of behaviour; but to record
self-reported behaviour. It can be argued that this would result in the
true incidence being under-reported as individuals are to some extent
practising self deception when they engage in misconduct.
Subsequently in order to preserve this deceit they will deny, even
anonymously, that they have done so. Conversely it can be argued that
males are more subject to anti social behaviour and bravado and so
might over report the level of their misconduct.
Since that time the questionnaire study has remained the prevalent
technique for assessing the level of misconduct in student cohorts
(Marsden et al 2005). Some have asked students to report the level of
behaviour in their cohort rather than their own (Sheard & Martin 2003)
in the hope that this would yield more accurate data. Other studies
Page 6
asked students to respond to hypothetical scenarios (Underwood
2003). While some asked tutors to report the amount of misconduct
(Culwin et al 2002) although tutor’s perceptions have been shown to be
an under estimate (Franklyn-Stokes & Newstead 1995).
The reported rates of misconduct vary widely between studies
(Stubbings & Brine 2003, Dick et al 2003, Hart 2004) which adds
confusion to attempts to state if the amount of misconduct is increasing
or decreasing. This confusion is caused by differences in the phrasing
of the questions, differences in the implicit institutional or explicit
questionnaire definitions of the terms used, differences in the way
measurements were made (Likert, nominal, interval) and differences in
the gross demographics of the cohorts studied (discipline area, gender
balances, age profiles, level of study). Hence although it is often stated
that the amount of misconduct is increasing there is very little, if any,
wide ranging and longitudinal evidence that this is the case. What
undoubtedly is the case is that the level of academic and stakeholder
interest has increased dramatically over the last decade.
The studies have generally reported that the level of reported
misconduct varies with gender, with males reporting higher levels than
females; and age, with younger students reporting more misconduct
than older students (Whitley et al 1999, Franklyn-Stokes. and
Newstead 1995, Hart 2004, Marsden et al 2005, Simon et al. 2004).
Other demographic factors are less clear-cut but, there is support for
the suggestion that misconduct is associated with lower levels of ability
and with students who have an instrumental attitude towards higher
education (Hart 2004, McCabe & Trevino 1997).
Page 7
The alternative to asking respondents to comment or report upon the
level of misconduct is to make a direct measurement of non-originality
within a corpus of collected student work. This method of measurement
can be capricious in at least two ways. Firstly the reported level of non-
originality will be lower than the true level. A study by Satterwhite &
Gerein (Satterwhite & Gerein 2001) took samples from 146 different
publicly available sources and submitted them to a number of different
detection services and search engines. The various services returned
between 20% and 58% hits, although the extent of detection that was
classified as a hit is not clearly stated. Secondly the tools report only
the extent of detected non-originality, this will include legitimately cited
material as well as illicitly reused material. However given these
caveats the levels of non-originality reported from such studies can be
regarded as a low water mark for the true extent of non-originality in a
corpus.
Other methodological problems relating to using direct measurement in
this way include the stability of the tools used and the meaning of the
values returned. Many of the tools are not stable and their performance
is continually being upgraded in the light of experience. This together
with the considerations that the tools have not been available for a
significant amount of time and that some that once existed no longer
appear to be available, makes longitudinal studies of cohorts or cohorts
passing through an institution problematic.
Page 8
In addition the meaning of the metric returned by the tool is not always
obvious or comparable, or in some cases even explained. It might be
assumed that a non-originality metric of 100% between two documents
would indicate that they were word for word identical. However some
tools, including the one described below, are fuzzy and a reported
metric of 100% does not mean that the documents are necessarily
totally identical. In a similar way a reported metric of 0% does not
mean that the documents had no words or phrases in common.
Accordingly direct comparisons between studies that use different
tools, or even the same tool on different occasions, may not be
possible.
Despite these problems there are a small number of studies that
attempt to report upon direct measurement of behaviour as expressed
in the proportion of detected non-originality in a submission. Many of
these studies report upon computer program source code plagiarism
as these tools have been available for a longer period of time and their
operation is simpler than those that are able to measure free text non-
originality (Chen et al 2002, Byrne et al 2004, Vamplew & Dermoudy
2005,). A smaller number of studies have investigated free text student
submissions (Knight et al 2004, Johnson et al 2004, Weinstein &
Dobkin 2002) and a related study by Clough (Clough 2001)
investigated non-originality in a corpus of press articles.
With the exception of the Weinstein study the papers are largely more
concerned with describing the operation, tuning and deployment of the
tools and only incidentally report upon the extent of the non-originality
detected and none relate it to any demographic factors. The Weinstein
Page 9
study is notable as it manually filtered out false hits by visual inspection
but again attempted no demographic analysis of the data obtained. It
did however present objective evidence that explicitly warning students
that their work would be subject to non-originality analysis caused the
level of non-attributed non-original content to fall.
The Weinstein study reports the underlying Internet plagiarism rate at
17% and claims that this is comparable with other, uncited, studies. It
divides these into 7% for ‘large scale plagiarism’ and 10% for
‘plagiarism small in magnitude’. However, the percentage non-
originality value, as provided by TurnItIn (TurnItIn 2005), upon which
this distinction is made is not stated. The Knight study is largely
concerned with optimising automated Google searches (Google 2005)
and reported 4 possible hits from 480 submissions. The Johnson study
used CopyCatch (CopyCatch 2005) and reported 4 plagiarised papers
from a corpus of 590 submissions.
It is against this background that the results obtained from this study of
direct measurement were related to demographic considerations in an
attempt to triangulate the general conclusions of the questionnaire
studies.
The Activity
The 2003/4 activity was designed following the experience in the
2002/3 session. This is a description of the latter activity with significant
changes from the previous activity commented upon. The activity
formed part of a level one professional skills unit taken by all first year
Page 10
computing and IT students and which is perceived by the staff as an
extended induction to higher education. A part of the unit involved a
lecture on the history of computing, which was followed up with a
homework task to write a related essay. A list of about 200 topics taken
from the history of computing had been prepared, allowing each
student to have an individualised task; which is also in line with good
practice guidelines. The instructions for the task defined the outputs
that the student had to produce which were: a title, some keywords, a
1000 word essay and two URLs where further information could be
obtained. The instructions advised the students to use World Wide
Web search engines to locate sources and included the explicit advice:
Although you are being encouraged to use the Web to
search for information you *MUST* consult several sources
of information and the essay you submit *MUST* be entirely
in your own words drawing only facts and ideas from your
sources.
The students were allowed two weeks to complete the activity and
were advised that they could contact the activities’ co-ordinator if they
could not locate any suitable material. A small number of students
made use of this support, most of whom needed assistance with
constructing effective search terms but some of whom had been given
a topic that was too obscure and needed to have a different topic
allocated.
The students were required to submit their essays and other outputs
using a Web form. Their essays were then made visible on the
University Intranet, indexed in various ways and students were
Page 11
encouraged to read the essays that other members of their tutor group
had produced.
Once the deadline for submission had passed the essays were
analysed for non-originality using a suite of programs based upon
OrCheck (Lancaster & Culwin 2004) technology. Essentially this
technology allows automated Google searches to be made, downloads
the documents indicated by the search engine and analyses the
degree of similarity between each downloaded document and the
document under investigation. The results of the analysis include a
visualisation that shows which parts of each located document have
been shown to be similar to parts of the document under investigation.
Copies of the target and located documents with the similarities
highlighted in different colours are also produced.
The results of the analysis were communicated to each student by
means of a pro-forma report which included the address of a hidden
URL, where the student’s submission and the most similar document
were shown side by side with the essentially identical segments
highlighted in red. An example of the web page produced for each
student is illustrated in Figure 1.
These results were made available to the students at the first tutorial
following a lecture on academic misconduct. That lecture illustrated the
process of academic misconduct by the lecturer pretending that they
had been required to write a 1000 essay on a historical figure called
"Tony Blair". The first essay was prepared by downloading the
biography of the prime minister from the official Number 10 web site,
measuring about 1000 words in Microsoft Word and inserting the name
Page 12
of the lecturer as the author. This essay was then analysed live by the
OrCheck tool with the source readily identified. A number of other pre-
prepared essays were then produced exhibiting different attempts at
disguise with each being submitted live to the tool with the sources
again being readily identified and the visualisation showing exactly how
each source had been used to produce the finished essay.
At this point the students were told that their essays had been
processed by the tool and that their tutors would be giving them the
results of the analysis and discussing the implications with them at the
next tutorial. It was also readily admitted that the tool was not foolproof
and it was quite possible that the headline reported percentage non-
originality value was lower, or even much lower, that the true degree of
non-originality. The lecture then continued by explaining what other
activities constituted academic misconduct, how and why these
activities damaged both the individual and, via the damage to the
institution, damaged everyone. The lecture concluded by outlining the
processes that would be followed if academic misconduct was
suspected by a tutor.
Page 13
Figure 1 On-line non-originality report
Page 14
There was no formal evaluation of the activity as such by the students.
However most of the tutors commented upon how amused and
engaged the students were, even by those who had unseemly
amounts of non-originality detected. A follow-up activity was planned
which involved a structured interview with a sample of the students,
with the sample biased towards those who had shown large amounts
of non-originality. However due to a combination of staff illness and
administrative confusion this activity did not take place.
The gross measurements
Figure 2 shows the cumulative non-originality graphs of the amount of
detected non-originality for the 145 students who participated in the
2002/3 activity and the 207 who participated in 2003/4.
Figure 2 Cumulative non-originality 2002/3 left and 2003/4 right
0
20
40
60
80
100
0 20 40 60 80 100
0
20
40
60
80
100
0 20 40 60 80 100
On each graph the x axis is the percentage of students who had at
least the degree of non-originality shown on the y axis. For example in
2002/3 20% of students had approximately 40% or more non-originality
in their essays. A small adjustment was made in the process of
Page 15
collecting and analysing for 2003/4 which had the effect of making the
process a little more sensitive at the lower end of the scale. Allowing
for this the graphs can be considered very similar. The mean in 2002/3
was 21.2% with a large standard deviation of 25.6%; in 2003/4 the
mean was 24.7% and coincidentally the standard deviation was also
24.7%. A t test confirmed that there was no significant difference
between the data from the two years.
When percentage non-originality values are stated in studies it is not
always clear how the value was obtained. The meaning of 100% non-
originality seems intuitively clear, all of the document can be shown to
be identical with parts of one or more other documents. The meaning
of 0% is less clear, naively it might be taken to indicate that there is no
similarity between the document under investigation and any other
documents considered in the analysis. However as all of the
documents are written in the same natural language there must be
some overlap in common vocabulary, even before any technical
vocabulary issues are considered. Values intermediate between 0%
and 100% are correspondingly even less clear.
The values reported here are computed by first removing the 200 most
common words in everyday English as reported by the British National
Corpus (BNC 2005) from all documents. The first word in the resulting
target document is then considered and the longest sequential
matching sequence of words starting with that word in any other
document is located. If that word does not exist in any other document
it scores 0 and the next word is considered. If the longest matching
string is two words, each word scores 0.25; three words score 0.5; four
Page 16
words 0.75 and five words or more 1.0. After scoring the sequence the
first word following it is considered. This process continues until all the
words in the target document have been scored.
Accordingly each word in the document can score between 0.0 and 1.0
with the gross reported percentage being the average score of all non-
common words in the document. The scoring system will not be unduly
influenced by a common technical vocabulary as terms or short
phrases, score lower that longer phrases or sections of matching text.
As the 200 most common words are removed before scoring it is
possible that a score of 100% could be obtained between the
document under investigation and a source or sources, despite them
not being word for word identical. This fuzzy matching of documents is
capable of seeing through some attempted disguise of the source
material by the plagiarist.
What was somewhat surprising was the number of students who
submitted a URL which subsequently proved to be the most significant
match. A total of 136 of the 207 students were shown to have more
than 10% non originality, a fairly arbitrary cut off value to exclude the
possibility of a common technical vocabulary and other non-significant
matching phrases. Of these 136, 61 that is approximately half,
submitted a URL that proved to be the greatest contributor of non-
originality. The average non-originality for these contributions was
Page 17
28.8% with a standard deviation of 16.4, the minimum was 11% and
the maximum 73%. The cumulative frequency graph for this group is
shown in Figure 3.
Figure 3 Cumulative non-originality attributable to a supplied URL
0
20
40
60
80
0 20 40 60 80 100
The graph indicates that about 40% of this group, about 54 students,
supplied a URL which contributed 25% or more of their essay; with
about 10% of the group, about 14 students, supplying a URL which
contributed 60% or more. The implications of this are either that the
students did not think that anyone would read their essays in
conjunction with the link supplied and note the similarity or that, despite
the clear injunction, they did not believe that this behaviour was
unacceptable.
The Demographic Analysis
Accepting the limitations of the measurement process the gross degree
of non-originality as observed can be used to investigate demographic
and other divisions within the sample. As the terms were randomly
Page 18
distributed through the sample the efficacy of the non-originality
measurement can be assumed to be non-capricious with respect
demographic and other variables. Consequently any observed
differences between sub-groups within the sample can be safely
attributed to membership of that group, rather than being an artefact of
the measuring process.
Gender
Table 1 shows the overall mean level of detected non-originality for
males and females. The left hand table sows the data from the 2002/3
session and the right hand table the data from the 2003/4 session. The
first row of the table shows the number of males and females, the
second row shows the level of non-originality for the entire sample and
the third row the level of non-originality excluding those who had less
than 10% non-original material detected.
Table 1. Analysis by gender 2002/3 left and 2003/4 right
female male female male
n 36 119 67 138
all 15.1 17.6 22.3 26.2
10 plus 17.1 20.5 35.3 36.7
Standard socio-biological theory states that as females are more risk
averse and general educational studies suggests that females are
more committed and organized. Both of these factors might suggest
that male non-originality would be higher than female non-originality,
and this is indicated by the data. However the difference was not
shown to be significant.
Page 19
Age
Figure 4 shows the relationship between age and non-originality. There
is no clear pattern in the graphs and the correlation coefficients
indicated no systematic relationship.
Figure 4 Scattergraph of age against non-originality 2002/3 left and
2003/4 right
0
20
40
60
80
100
15 25 35 45 55
age
non-originality
0
20
40
60
80
100
15 20 25 30 35 40 45 50 55
age
non-originality
When the sample was divided into those younger than 22 and those
older than 22, a t test showed no significant differences. The value 22
was chosen as is about the minimum age for individuals to graduate in
UK universities. As with gender it might be hypothesised that older
students would be less likely to engage in misconduct and although
this is indicated by the raw data it cannot be concluded. The data is
presented in Table 2.
Table 2. Analysis by age 2002/3 left and 2003/4 right
<22 >=22 <22 >=22
n 83 73 101 103
all 18.9 15.4 27.0 22.8
10 plus 28.1 23.4 34.9 38.8
Page 20
Outcome
Outcome can be measured in two ways. The students submitted the
essay in November with the end of the academic year the following
June. Some students withdrew before completing the year and so
would have an incomplete set of results. The remaining students'
outcome could be measured by taking the overall percentage score as
recorded on their first year transcript.
Table 3. Analysis by completion 2002/3 left and 2003/4 right
complete withdrawn complete withdrawn
n 129 27 182 25
all 16.3 21.9 23.8 31.3
10 plus 24.3 34.2 34.8 45.0
Table 3 shows the average non-originality scores of the minority of
students who did not complete their first year of study and of those who
did. A t test on the entire cohorts showed the difference to be
significant 0.1 confidence level for the whole groups in both years
(2002/3 p=0.083, 2003/4 p=0.078). When the analysis was restricted to
those with more than 10% non-originality the differences for both years
were significant at the 0.05 confidence level (2002/3 p=0.033, 2003/4
p=0.047).
The data suggests that the amount of non-originality is indicative of not
completing the year. That is those first year students who submitted
higher levels of non-original material in November were significantly
less likely to present a full set of marks to the examination board the
following July. This result does not seem surprising as students who
have to resort to cheating so early in their higher education academic
Page 21
progression would seem to be ill prepared for the demands of studying
at this level. It might also be that they are less committed to studying
and so are more likely to withdraw.
Figure 5 presents scattergraphs of the students who completed the
year against non-originality. There is no clear pattern in the graphs and
the correlation coefficient indicated no systematic relationship.
Figure 5. Scattergraph of outcome against non-originality
0
20
40
60
80
100
0 20 40 60 80
outcome
non-originality
0
20
40
60
80
100
0 20 40 60 80outcome
non-originality
When the sample was divided into those whose outcome was less than
40% and those above 40%. The 2002/3 averages were 20.4% and
15.5% respectively, a difference which was not significant. The 2003/4
averages were 25.9 and 15.8, a difference which was shown by a t-test
to be significant at the 0.05 level (p=0.042). The value of 40% was
chosen as this is regarded as the pass/fail boundary at LSBU.
Following on from the non-completion argument it would appear that
there is some evidence that weaker students are more prone to
engage in misconduct.
Page 22
Course
The students in the sample came from one of two programmes:
Computing, which includes both computing and internet computing and
Business Information Technology (BIT), which includes both BIT and e-
commerce. The computing programme is regarded as the more
technical and students entering it tend to have a higher level of
qualification. Table 4 shows the non-originality related to course, the
number of students in this analysis is smaller than that in previous
tables as the course data for a number of students was clearly
erroneous. A t test on the 2002/3 data showed significance at the 0.05
level, both for the entire sample and for those with more than 10% non-
originality (t=1.74, t crit=1.66, p=0.042 & t=1.89, t crit=1.66, p=0.031).
The data from the 2003/4 cohort was non-significant.
Table 4. Analysis by course 2002/3 left and 2003/4 right
Computing
BIT Computing
BIT
n 64 84 84 115
all 14.0 19.5 25.9 22.9
ten plus
21.9 29.4 33.9 37.6
A tentative conclusion from the 2002/3 data might be that there is
some evidence that less technical courses are associated with higher
Page 23
levels of misconduct. This conclusion would be in keeping with the
amount of plagiarism noted on final year projects that are organised
and administered identically across both programmes. However it was
not possible to triangulate this conclusion from the 2003/4 data.
Triangulation with MOSS
In the 2002/3 session the computing students were required to
complete a Java programming coursework; in 2003/4 all students were
required to do so. The program listings were collected in both years
and submitted to the Measure Of Software Similarity (MOSS 2005)
service, which measures the extent of intra-corporal similarity in the
source code listings. This provides an independent measure of non-
original material and so can be triangulated with the data from the
essay activity. The scattergraphs of the 24 students from the 2002/3
cohort and the 69 students from the 2003/4 cohort, whose free text
non-originality was greater than 10% and for whom MOSS measures
were available are given in Figure 5.
Figure 5. Scattergraph of souce code against free text non-originality,
2002/3 left and 2003/4 right
0
10
20
30
40
50
60
70
0 10 20 30 40 50 60 70 80 90 100
source code
free text
0
10
20
30
40
50
60
70
80
90
0 20 40 60 80 100 120
source code
free text
Page 24
Although the correlation from the 2002/3 data was significant at the
0.01 level, the correlation from the 2002/3 data was non-significant.
The implication from the significant correlation would have been that
presenting non-original work appears to be consistent between
subjects and modalities, however this was not confirmed from the
larger 2003/4 sample.
Conclusions and further work
This paper gives an example of how first year undergraduate students
can be actively introduced to the issues of academic misconduct.
Although no formal evaluation of the activity was conducted the reports
from the tutors and from the students themselves would seem to
indicate that this activity was more favourably received than a more
passive exposition which was presented in previous years.
The study also shows that it is possible to systematically measure the
amount of non-originality in student submissions and from that
investigate the demographics of academic misconduct. The relatively
short length of the student submissions in this study and the somewhat
artificial nature of the task may have affected the outcome. It is also
possible that an institutional memory is in the process of being
developed whereby first year students are pre-warned that the history
essays will be subject to non-originality investigation. If this is the case
then the amount of non-originality contained within the submissions
would decline; however this is exactly what is wanted from a
programme to educate students about academic misconduct.
Page 25
Several factors which have been suggested as possibly significant,
were indicated but were not shown to be significant. The only
conclusion that is as yet unambiguously concluded is that students
who subsequently did not complete their first year were more likely to
present non-original material. This is partly coupled with the conclusion
that weaker students are significantly more likely to engage in
cheating. If these findings can be shown to be replicable then it would
indicate that activities such as this can be used to identify students who
might benefit from additional support. It would be worthwhile for the
study to be repeated in subject areas other than computing and in
institutions other than LSBU. Not only would this further triangulate the
data, if common measurement techniques were applied, would allow
the demographic influences of broad subject area and type of
institution to be investigated.
Page 26
References
Barbrook, R. (2002), The Napsterization of Everything, Science as Culture, 11(2) 277-285.
Boling Elizabeth & Frick Theodore (undated), How to Recognize Plagiarism, Indiana University, School of Education, available at http://ww w.indiana.edu/~istd/practice.html (accessed 1 April 2005).
BNC (2005), British National Corpus, available at www.natcorp.ox.ac.uk (accessed 1 April 2005).
Byrne Siobhán, Charles Daly, Jane Horgan & John Waldron (2004), A technique for measuring plagiarism in automated & distance learning environments, Proceedings of the winter international symposium on Information and communication technologies, Mexico 2004
Carroll Jude & Appleton Jon (2001), Plagiarism A Good Practice Guide, Report commissioned by the Joint Information Systems Committee (JISC) available at http://online.northumbria.ac.uk/faculties/art/information_studies/Imri/Jiscpas/docs/brookes/brookes.pdf (accessed 1 April 2005).
Carroll Jude (2002), A Handbook for Deterring Plagiarism in Higher Education, Oxford Brookes University, ISBN 1873576560
Chen Xin, Li Ming, Mckinnon Brian & Seke Amit (2002),A Theory of Uncheatable Program Plagiarism Detection and Its Practical Implementation, available at www.cs.ucsb.edu/~mli/sid.ps, (accessed 1 April 2005).
Clough Paul (2001), Measuring text reuse in a journalistic domain, Proceedings of the 4th Annual CLUK Colloquium, Sheffield, UK, 53-63
Page 27
CopyCatch (2005), http://www.copycatchgold.com, (accessed 1 April 2005).
Culwin, Fintan & Lancaster Thomas (2001), Plagiarism, prevention, deterrence and detection, report commissioned by the Instutute for Learning & Teaching (ILT) available at http://www.heacademy.ac.uk/embedded_object.asp?id=21635 (accessed 1 April 2005).
Culwin, F., MacLeod, A. & Lancaster, T. (2002), Source Code Plagiarism in UK HE Computing Schools: Issues, Attitudes and Tools, South Bank University Technical Report SBU-CISM-01-02.
Culwin Fintan (2005), PlagPoster available at http://cise.lsbu.ac.uk/plagposter (accessed 1 April 2005).
Dick Martin, Sheard Judy, Bareiss Cathy, Carter Janet, Joyce Donald, Harding Trevor, Laxer Cary (2003), Addressing student cheating: definitions and solutions, ACM SIGCSE Bulletin 35(2) 172-184
Franklyn-Stokes,A. and Newstead,S. (1995), Undergraduate cheating: who does what and why?, Studies in Higher Education, 20(2), 159-172.
Goolge (2005), http://www.google.com/apis/ (accessed 1 April 2005).
Harris Robert (2004), Anti-Plagiarism Strategies for Research Papers, available at http://www.virtualsalt.com/antiplag.htm (accessed 1 April 2005).
Hart Mike (2004), Plagiarism and Poor Academic Practice – A Threat to the Extension of e-Learning in Higher Education?, Electronic Journal
Page 28
of E-Learning (EJEL) 2 (2), available at http://www.ejel.org/volume-2/vol2-issue1/issue1-art25.htm 2004 (accessed 1 April 2005).
Johnson D, Patton R, Bimber B, Almeroth K & Michaels G (2004), Technology and Plagiarism in the University: Brief Report of a Trial in Detecting Cheating, Association for the Advancement of Computing in Education (AACE) Journal, 12(3) 281-299.
Knight Alan, Almeroth Kevin & Bimber Bruce (2004), An Automated System for Plagiarism Detection Using the Internet, World Conference on Educational Multimedia, Hypermedia & Telecommunications (ED MEDIA), Switzerland 2004.
Lancaster Thomas & Culwin Fintan, Using freely available tools to produce a partially automated plagiarism detection process, Proc ASCILITE 2004 pp520-529, available at http://www.ascilite.org.au/conferences/perth04/procs/pdf/lancaster.pdf (accessed 1 April 2005).
McCabe Donald L. & Trevino Linda Klebe (1997), Individual and Contextual Influences on Academic Dishonesty: A Multicampus Investigation,Research in Higher Education, 38(3) 379 - 396
Marsden Helen, Carroll Marie & Neill James T, Who cheats at university? A self-report study of dishonest academic behaviours in a sample of Australian university students, Australian Journal of Psychology, 57(1) 1-10
MOSS (2005), Measure Of Software Similarity, available at http://www.cs.berkeley.edu/~aiken/moss.html (accessed 1 April 2005).
Ross, George MacDonald (2004), Plagiarism really is a crime: a counterblast against anarchists and postmodernists, Proc. Plagiarism:
Page 29
Prevention, Practice & Policy conference, Northumbria University Press, 69-70
Satterwhite, Robin & Marla Gerein. (2001) Downloading detectives: searching for on-line plagiarism, available at http://www.coloradocollege.edu/library/course/webplag2.html (accessed 1 April 2005).
Sheard Judy & Dick Martin, Influences on cheating practice of graduate students in IT courses: what are the factors?, ACM SIGCSE Bulletin, 35(3) 45-49
Simon Christopher A., Carr Jim R., Mccullough Sesi M., Morgan Sally Oleson J., Ted & Ressel Maggie (2004), Gender, student perceptions, institutional commitments and academic dishonesty: who reports in academic dishonesty cases?, Assessment & Evaluation in Higher Education 29(1) 75-90
Stubbings Ruth & Brine Alan (2003), Plagiarism and the role of the library, Update (magazine of the Chartered Institute of Library and Information Professionals) Issue December 2003 available at http://www.cilip.org.uk/publications/updatemagazine/archive/archive2003/december/update0312c.htm (accessed 1 April 2005).
Szondi Juliet & Martindale Chris (2004), PLATO: PLagiarism Teaching Online available at http://www.derby.ac.uk/focus-on/archive/internal.asp (accessed 1 April 2005).
Turnitin, http://www.turnitin.com (accessed 1 April 2005).
Underwood Jean D. M. (2003), Student attitudes towards socially acceptable and unacceptable group working practices, British Journal of Psychology 94(3), 319-317
Page 30
Vamplew, P.W. & Dermoudy, J.(2005), An Anti-Plagiarism Editor for Software Development Courses, Proceedings of the Australasian Computing Education Conference, Newcastle, Australia
Weinstein J.W. & Dobkin CE, Plagiarism in US Higher Education: Estimating Internet Plagiarism Rates and Testing a Means of Detterence, available at http://webdisk.berkeley.edu/~Weinstein/Weinstein-JobMarketPaper.pdf (accessed 1 April 2005).
Weller Martin J (2003), Suggested activities for students, Open University, available at http://online.northumbria.ac.uk/faculties/art/information_studies/Imri/Jiscpas/site/adv_academicprac.asp (accessed 1 April 2005).
Whitley Bernard E., Bichlmeier-Nelson Amanda and Jones Curtis J. (1999) Gender Differences in Cheating Attitudes and Classroom Cheating Behavior: A Meta-Analysis, Sex Roles 41(9) 657 - 680